The Levin Report and SK&A data are both proprietary sources, and as such, we cannot widely disseminate them. We are able to share them with researchers who have already purchased both data sets. Please reach out to the corresponding author if you have interest in our versions of the Levin Reports and SKA data relevant to our analysis. The code provided in this replication package starts with code/data that presumes the existence of the intermediate data to that point. 

We include 8 .dta data sets: chronic_3digit.dta, chronic_4digit.dta, diab_5digit.dta, hyp_3digit.dta, hyp4_digit.dta, and hyp_5digit.dta. These merge onto the claims data to identify those claims associated with the various intensity levels of the chronic conditions we consider.

We also include an excel spreadsheet, X_BalanceStats.xlsx, since we use it to construct the normalized differences formula. 

Program Setup: We have 6 folders with numbered prefixes. Each folder includes Stata and R programs that create data and generate results. The programs are to be run progressively, such that the programs in 1 are intended to be run prior to the programs in 2, etc. Similarly, the programs in each folder are also numbered and are intended to be run progressively.

Folders:
1.Create Data - These programs create a quarterly dataset of beneficiaries from the Medicare claims and merge on the acquisition and health information.
2.Analyses Matching - These programs use the estimated propensity score to create the matched sample and estimate the acquisition effects among the matched sample.
3.Analysis_Parallel Trends- These programs create the parallel trend graphs for the matched sample 
4.Analysis_Full Sample OLS - These programs estimate the linear probability model acquisition effect estimates among the full sample. They also generate the mean statistics.
5.Analysis_Blocking - These programs create the blocks from the propensity scores and estimate the acquisition effects using the blocking analysis.
6.Create Tables and Figures- The programs in this folder use the output generated from the other programs to create the tables and figures included in the paper.

Each .do file has a set of macros that creates the file path for the data and called programs. Those will need to be changed for the programs to work. Similarly, the R programs have paths that will need to be changed. Each program provides a list of the datasets used and the datasets created. Ensure that the relevant paths contain the relevant files for the program to be run. 

We do not include all of the programs or effort used to generate the list of of NPIs(UPINs) identified in the mergers since they come from the proprietary SK&A and Levin Report data. 


